1548
Energy & Fuels 2005, 19, 1548-1560
Molecular Weight of Petroleum Asphaltenes: A
Comparison between Mass Spectrometry and Vapor
Pressure Osmometry
Sócrates Acevedo,* Luis B. Gutierrez, Gabriel Negrin, and Juan Carlos Pereira†
Laboratorio de Fisicoquı́mica de Hidrocarburos, Escuela de Quı́mica, Facultad de Ciencias,
Universidad Central de Venezuela, 47102, Caracas 1041, Venezuela
Bernardo Mendez
Laboratorio de Resonancia Magnética Nuclear, Escuela de Quı́mica, Facultad de Ciencias,
Universidad Central de Venezuela, 47102, Caracas 1041, Venezuela
Frederic Delolme and Guy Dessalces
Service Central d’Analyses du CNRS, BP 22 69390 Vernaison, France
Daniel Broseta
Laboratoire des Fluides Complexes, Université de Pau, BP 1155, 64013 Pau, France
Received August 6, 2004. Revised Manuscript Received February 22, 2005
Seven asphaltene samples and six octylated asphaltene (OA) derivatives were analyzed using
laser desorption ionization-time-of-flight (LDI-TOF) mass spectrometry (abbreviated as MS)
and vapor pressure osmometry (VPO) techniques. Molecular weight distributions (MWDs) that
were determined using MS spanned, for all asphaltenes samples and their octylated counterparts,
a similar range, from ∼100 Da to ∼10 000 Da. For asphaltenes, the number-average molecular
weight (Mn) and weight-average molecular weight (Mw) afforded values in the 1900 ( 200 and
3200 ( 400 intervals, respectively. Consistently heavier values were observed for the OA
derivatives (Mn ≈ 2300 ( 200 and Mw ≈ 3600 ( 200). To select the adequate laser power for
these measurements, experiments at different laser powers were performed, to increase volatility
and reduce fragmentation to a minimum. Several other experiments were performed to validate
these results. First, good agreement between the measured and calculated Mn values was observed
for OA materials (calculated from asphaltene Mn and n, the number of octyl groups introduced,
as determined from elemental analysis); second, Mn values, as measured by VPO and MS, were
determined to be equal, within an average standard deviation of (27.0%. These results and
calculations strongly suggest that the MWD, the molecular weight range, and the molecular weight
averages determined using the present MS technique are reasonable estimates of the molecular
weight properties of asphaltenes and not the result of artifacts such as fragmentation,
polymerization, incomplete volatilization, etc., which may be occurring during the MS experiment.
1. Introduction
Determination of the molecular-weight-related properties of asphaltenes is an important step toward the
characterization of these substances, and, thus, much
effort has been devoted to this goal.1-20,24,25 However,
* Author to whom correspondence should be addressed. Telephone:
58 212 6051246/58 212 6051218. Fax: 58 212 3725017. E-mail
address: [email protected].
† Current address: Universidad de Carabobo, Facultad Experimental de Ciencia y Tecnologa, Departamento de Quı́mica 3336, Valencia
2001, Caracas.
(1) Moschopedis, S. E.; Fryer, J. F.; Speight, J. G. Fuel 1976, 55,
227.
(2) Speight, J. G.; Wernick, D. L.; Gould, K. A.; Overfield, R. E.;
Rao, B. M. L. Rev. Inst. Fr. Pet. 1985, 40, 51.
(3) (a) Reerink, H.; Lijzenga, J. Anal. Chem. 1975, 47, 2160. (b)
Hayes, M. H. B.; Stacey, M.; Standley, J. Fuel 1972, 51, 27.
(4) De Canio, S. J.; Nero, V. P.; De Tar, M. M.; Storm, D. A. Fuel
1990, 69, 1233-1236.
(5) Winans, R. E.; Hunt, J. E. Prepr. Symp.sAm. Chem. Soc., Div.
Fuel Chem. 1999, 44 (4), 725.
because of several factors (such as polydispersity of the
sample), difficulties in isolation from crude oils, limitations of techniques used to measure the values, and the
tendency to form aggregates, such determinations have
(6) Miller, J. T.; Fisher, R. B.; Thiyagarajan, P.; Winans, R. E.; Hunt,
J. Energy Fuels 1998, 12, 1290.
(7) Boduszynsky, M. M. Energy Fuels 1987, 1, 2.
(8) Yang, M.; Eser, S. Prepr. Symp.sAm. Chem. Soc., Div. Fuel
Chem. 1999, 44 (4), 768.
(9) Sheu, E. Y.; Storm, D. A. Fuel 1994, 73, 1368.
(10) Leite, L. F.; Camillo, M. C. F.; Deane, G. H. W.; Brandao, L.;
Cintra, M.; Carvallo, J. R. F. J. High Resolut. Chromatogr. 1989, 12,
498.
(11) Groenzin, H.; Mullins, O. J. Phys. Chem. A 1999, 103, 11237.
(12) Groenzin, H.; Mullins, O. Energy Fuels 2000, 14, 677.
(13) Nali, M.; Manclossi, A. Fuel Sci. Technol. Int. 1995, 13, 1251.
(14) Yarranton, H. W.; Alboudwarej, H.; Jakher, R. Ind. Eng. Chem.
Res. 2000, 39, 2916.
(15) Acevedo, S.; Escobar, G.; Ranaudo, M. A.; Rizzo, A. Fuel 1998,
77, 853.
(16) Acevedo, S.; Escobar, G.; Gutierrez, L. B.; D’Aquino, J. A. Fuel
1992, 71, 1077.
10.1021/ef040071+ CCC: $30.25 © 2005 American Chemical Society
Published on Web 06/14/2005
Molecular Weight of Petroleum Asphaltenes
been proven to be very difficult. Thus, it is not surprising
that molecular weights between 300 000 and 1000 have
been reported in the past.1,2 It is now recognized that
the major source of these discrepancies is the formation
of aggregates or macrostructures formed through intermolecular interactions.
Vapor pressure osmometry (VPO),1,2,14-16,20,24 size
exclusion chromatography (SEC),3,10,14-16,20 and mass
spectrometry (MS)4-9,17,18,24 are among the techniques
most commonly used for molecular weight studies of
asphaltenes and bitumen. Because of its simplicity and
cost, VPO is used extensively. The VPO technique
allows for the determination of an “absolute” value for
the number-average molecular weight (Mn). When used
for weakly or nonpolar low-molecular-weight compounds, such as resins and crude oils, the technique
gives accurate values, similar to those obtained by mass
spectroscopy.8 However, aggregate formation, which is
a factor to consider with asphaltenes, is always difficult
to eliminate completely, even with the use of high
temperatures and polar solvents.1,2 Molecular weight
distributions (MWDs) as well as the Mn and the weightaverage molecular weight (Mw) for asphaltenes has been
obtained using SEC.3,16,19 However, complications that
are due to inadequate calibration, flow rate effects,
column adsorption, and detection of sample severely
limits the application to asphaltenes. Moreover, being
a relative technique, calibration with adequate standards is required and special methods must be used to
obtain meaningful results.16
Satisfactory fittings to a log-normal distribution
(LND) have been found for asphaltenes and octylated
asphaltene (OA) derivatives when using SEC techniques.3,15 This strongly suggests that compounds in the
mixture are the result of a natural statistical process
that occurs during asphaltene generation (probably a
cracking of giant molecules). From a mathematical
standpoint, polymerization and depolymerization are
equivalent and the situation could be compare to a usual
polymerization leading to MWDs similar to LNDs.21
Using thermal volatilization (from 50 °C to 300 °C),
which is a low-voltage, low-resolution mass spectroscopy
technique, molecular weights of ∼400 have been reported.5 As a result from laser desorption mass spectra
(LD MS) techniques, it has been claimed that asphaltene molecular weights should not be higher than 1500,
with average values of ∼400.6,7 Similar low values, using
similar mass spectra techniques were reported previously7 and more recently.8 Molecular weight values in
the range of 1000 or less have been reported using
(17) Domin, M.; Herod, A.; Kandayoti, R.; Larsen, J. W.; Lazaro,
M.-J.; Li, S.; Rahimi, P. Energy Fuels 1999, 13, 552.
(18) Larsen, J. W.; Li, S. Energy Fuels 1995, 9, 760.
(19) Acevedo, S.; Mendez, B.; Rojas, A.; Layrisse, I.; Rivas, H. Fuel
1985, 64, 1741.
(20) Gutierrez, L. B.; Ranaudo, M. A.; Méndez, B.; Acevedo, S.
Energy Fuels 2001, 15, 624.
(21) Billmeyer, F. W., Jr., Ed. Textbook of Polymer Science; WileyInterscience: New York, 1970; Chapters 1, 8, and 9.
(22) Hillenkamp, F.; Karas, M.; Beavis, R. C.; Chait, B. T. Anal.
Chem. 1991, 63, 1193A.
(23) Hoffmann, E.; Stroobant, V. Mass Spectrometry. Principles and
Applications, Second Edition; Wiley: New York, 2002, Chapter 1, p
29.
(24) Tanaka, R.; Sato, S.; Takanohashi, T.; Hunt, J. E.; Winans, R.
E. Energy Fuels 2004, 18, 1405.
(25) Ignasiak, T.; Kemp-Jones, A. V.; Strausz, O. P. J. Org. Chem.
1977, 42, 312-320.
Energy & Fuels, Vol. 19, No. 4, 2005 1549
interfacial tension data9 and, more recently, using a
fluorescence depolarization technique.11,12
However, many other reports, using VPO, have
indicated that Mn values are higher and probably in the
range of 1500-2000.1,2,14,16 Use of polar solvents such
as nitrobenzene at high temperatures usually leads to
VPO values in that range (see below). Similarly, MW
ranges reaching high MW valuessfrom 3000 (from ref
18) to 10 000 (from ref 24) and 20 000 (from ref 17)s
have been reported using MS techniques.
It could be argued that some of the previously
mentioned discrepancies are mainly due to either
fragmentation, recombination, or incomplete volatilization of sample during the mass spectra analysis, or to
sample aggregation during the VPO measurements.
Volatilization is an issue that is particularly important
in regard to MS methods. For instance, experiments
with laser desorption show that the MWD obtained is
dependent on the laser power used. Because of incomplete volatilization, MWDs that have been measured at
low power are unreliable. On the other hand, excessive
power leads to fragmentation (see below). Thus, a
compromise is needed to obtain optimal conditions.
Because laser energies could be controlled to reduce
sample fragmentation to a minimum, matrix-assisted
laser desorption ionization-time-of-flight (MALDITOF) MS techniques are called “soft” MS techniques.
As is well-known, MALDI MS has been used extensively
to analyze biopolymers and synthetic polymers, as well
as polycyclic aromatic hydrocarbons (PAHs).22,23 Matrixes are used for samples such as proteins, peptides,
and carbohydrates, which do not strongly absorb the
laser energy. However, for absorbing high-molecularweight compounds, such as asphaltenes, matrixes are
not necessary (see below). This was determined to be
the case for the asphaltene samples analyzed by Tanaka
et al., using a LDI apparatus and techniques similar to
those used here24 (see below).
The possible reactions that could be occurring during
asphaltene octylation have been discussed earlier.25
Thus, only a brief and schematic account of these is
given here. When treated with the reagent potassium
naphthalenide, S atoms in thiophene-like structures
react according to the reaction given in Scheme 1:
Scheme 1. Model Reactions To Illustrate Asphaltene
Octylationa
a Here, the dibenzothiophene functional group I represents an
asphaltene molecular section. Potassium naphthalenide is represented by groups II and III, groups IV and V are intermediaries,
and group VI is the octylated product.
1550
Energy & Fuels, Vol. 19, No. 4, 2005
Scheme 2
There are two properties of OA worth mentioning at
this stage. Changing aromatic (thiophene-like) S-C
bonds to aliphatic bonds make OA materials more labile
than asphaltenes. For instance, as shown by (unpublished) thermogravimetric analysis (TGA) experiments,
OA materials loose up to 18% of their weight when the
sample was heated from room temperature to 350 °C
(compare with the 3% loss for asphaltenes) This occurs
in one band (centered at ∼250°) before sample pyrolysis,
which occurs at temperatures of ∼400° and above. This
should be due to the cracking of aliphatic C-S bonds.
Thus, should there be any significant fragmentation
reactions during LD MS experiments, these would show
clearly in the OA MS.
For similar reasons, it is expected that any possible
reaction, such as free-radical recombination, leading to
a polymeric species (see reaction Scheme 2), would be
more likely to occur with OA materials than with
asphaltenes. As shown below, any significant change in
the MWD of asphaltenes that is due to this or any other
related process could be excluded by the results obtained.
Another OA property is related to volatility and
melting points. As shown below, contrary to asphaltenes, which do not melt, OA materials have melting
points of ∼100 °C. Thus, it is expected that OA materials
would be volatile enough to set the adequate power
required to get all of the asphaltene sample in the vapor
phase during the MS measurements.
In this work, we present an MS study of 13 asphaltene samples using a LDI-TOF technique (no matrix
used) that is capable of analyzing molecular weights of
up to 30 000 Da and higher. The MS results obtained
were compared with VPO values. Similarly, the Mn
values for OA materials, measured using MS, were
compared to the calculated values. As described below,
the results of these comparisons and calculations lead
us to believe that the MS results are reliable and, hence,
that the averages and MW range of asphaltenes are
those referenced in the previously mentioned Abstract
section.
Experimental and Methods Section
2.1. Asphaltenes. Furrial, which is an oil with asphaltene
precipitation problems, was obtained from Monagas State in
the northeast part of Venezuela. DM 153, DM 115, and VLA
711 were from the Maracaibo Lake in northwestern Venezuela.
Hamaca and Cerro Negro were extra heavy crude oils from
the Orinoco basin in southwestern Venezuela. As described
previously,19 because of the high viscosity of these last oils,
they were diluted with toluene (1:1) before the addition of
n-heptane to obtain the asphaltenes. For the other samples,
no dilution with toluene was required. In all cases, the
precipitated solid was continuously extracted with boiling
n-heptane in a Soxhlet extractor, to remove coprecipitated
resins (usually 24 h).
A1 was a Furrial fraction isolated using the so-called PNP
method.20 In this method, a toluene solution of the asphaltene
is treated with para-nitrophenol (PNP) and, from this mixture,
a PNP-A1 complex is precipitated, from which A1 is separated
by alkaline treatment. The solubility of the A1 fraction in
toluene at room temperature is very low (0.09 g/L) when
compare to the asphaltene from which is isolated (∼57 g/L)
Acevedo et al.
2.2. Reductive Octylation: Synthesis of Octylated
Asphaltenes (OA). Octylated asphaltenes (OAs) were synthesized by treating the asphaltenes with potassium naphtenide in tetrahydrofuran (THF) and then with octyl iodide, as
reported elsewhere.16,25
2.3. 13C NMR Experiments. NMR spectra were recorded
on a JEOL model EX 270 Fourier transform NMR spectrometer using CDCl3; tetramethylsilane (TMS) was used as an
internal standard. The instrument was equipped with a 5-mm
broadband probe head. Processing was performed using the
DELTA V1.8 program, which was running on a Silicon
Graphics workstation. The selected parameters were as follows: spectral window, 250 ppm; width of 30° pulses, 2.8 µs;
relaxation delay, 5 s; and number of scans, 8000-12 000. The
spectra were obtained using standard JEOL software.
2.4. Number of Octyl Groups (n). The number of octyl
groups (n) (per 100 C atoms in the asphaltenes) incorporated
into the asphaltene molecule after the octylation treatment
was obtained using elemental analysis. For elemental analysis
and from the corresponding H/C ratios, using eq 1, as described
earlier:16,25
(H/C)O )
100(H/C)A + 17n
100 + 8n
(1)
or
n)
100((H/C)O - (H/C)A)
17 - 8(H/C)O
(2)
Expected errors in n were calculated using the procedure
described below for the Mn(OA) case.
2.5. Mass Spectra. LDI mass spectra were recorded in a
positive polarity, using an Applied Biosystems apparatus
(MALDI-TOF Voyager DE-STR). A N2 laser (λ ) 337 nm) was
used. Samples were dissolved in THF (10 g/L) and then 1 µL
of the solution was deposited dropwise on the target.
To minimize fragmentation that occurred with inappropriate
laser-power values, leading to noise and ions mainly in the
low mass range (typically m/z < 500), the power selected was
optimized both for OA and asphaltenes. The laser power used
(measured in terms of the number of laser shots (LS)) were
1900, 2100, 2300, and 2500 LS. For the OA materials, low
power (1900 LS) afforded reasonable spectra with little
fragmentation. However, spectra with low intensities and poor
resolution were obtained for asphaltenes under the same
conditions. The optimal power (2100 LS) was selected by
monitoring the change of the MWD of the OA material as the
laser power was increased. Several trials were performed using
a matrix (dithranol: 1,8,9-anthracenetriol). For this, asphaltenes and the matrix were dissolved in chloroform, the solvent
was evaporated, and the resulting solid (2% in asphaltenes)
was analyzed. No significant differences were obtained between these MALDI spectra and the LDI spectra measured
for the asphaltene without matrix. Therefore, no matrix was
used in the spectra reported here.
Moreover, all the spectra were recorded in the first few
minutes after the introduction of the sample in the mass
spectrometer, which avoided the fast volatilization of molecules
of low molecular weight, leading to a modification of the
sample’s distribution.
2.6. Estimation of Mn for Octylated Asphaltenes. These
were estimated from the Mn values of asphaltenes, using eq 3
below:
(
Mn(OA) ) 1 +
)
117n
M (A)
M100 n
(3)
Here, M100 is the MW per 100 C atoms, which is obtained from
the elemental analysis (∼1400 in all cases).
Molecular Weight of Petroleum Asphaltenes
Energy & Fuels, Vol. 19, No. 4, 2005 1551
Table 1. Elemental Analysis and Other Properties of Asphaltenes and Crude Oils
Elemental Analysis (%)a
sample
API of
crude oila
amount of asphaltene
in crude oil (%)
C
H
Ob
N
S
H/Cc
Furrial
DM 153
DM 115
VLA 711
Hamaca
Cerro Negro
21
14
11.2
31.1
8
8
7.6
10.0
11.0
2.2
14
13
85.50
81.05
80.23
83.94
83.94
81.23
6.90
8.17
7.73
7.53
7.71
7.72
2.5
2.0
4.4
2.0
1.9
3.5
1.73
1.64
1.44
1.23
1.96
2.13
3.4
6.6
6.3
4.6
4.5
5.5
0.97
1.20
1.10
1.06
1.10
1.12
a
In asphaltenes. b By difference. c With errors of ∼3% or less.
Expected relative errors in Mn(OA) were calculated using
the usual procedure. This leads to eq 4:
100 ×
dMn(OA)
Mn(OA)
Table 2. Elemental Analysis and Other Properties of
Octylated Asphaltenes
Elemental Analysis
)
100 ×
{
117dn
1400[1 + (117n/1400)]
+
}
dMn(A)
Mn(A)
sample
(4)
The parameter dMn(A) is unknown; therefore, we assumed it
to be equal to 0.2Mn(A).
2.7. VPO Measurements. With the described instrument,16
VPO Mn were measured in nitrobenzene at 100 °C using five
solutions in the 0-6 g/L range. Plots of ∆V (instrument
readings) against the concentration C were fitted to parabolas
such as those defined by eq 5.
∆V ) aC2 + bC + d
(6)
The Mn value was then obtained from eq 7, where K is the
instrument constant:
Mn )
K
b
75
80
70
85
70
70
(7)
Because eq 5 (and eq 6) contains three parameters, this
procedure is generally more accurate than the usual ∆V/C vs
C plot, which contains only two parameters (a different pair
of parameters a and b).
3. Results
As shown in Table 1, the asphaltene samples span a
significant range of H/C values and sulfur contents.
Also, the corresponding oils span a wide range of
asphaltenes content and API gravities from the extra
heavy (Hamaca and Cerro Negro) to the light VLA 711.
Therefore, these samples could be considered as a good
general representation of the oil and asphaltenes.
The elemental analysis and other properties of the
octylated derivatives are shown in Table 2. As expected,
the H/C ratios of these samples were higher than those
corresponding to asphaltenes. As shown in Table 2, all
octylated samples melt at <105 °C, with reasonable
melting ranges.
The 13C spectra of DM 153 samples are shown in
Figure 1. Similar spectra were measured for the other
samples, and the complete results are shown in Table
3. As expected, the percentages of aliphatic carbons were
higher for the octylated derivatives; n values in Table
3 were calculated using eq 2.
87-91
80-85
97-104
85-97
100-105
100-104
C
Ob
H
N
S
H/Cc
82.67 8.33 5.56 0.92 2.5 1.20
81.87 10.15 1.93 0.55 5.5 1.48
81.69 9.16 2.65 1.30 5.2 1.34
83.15 9.73 2.73 0.79 3.6 1.39
84.03 9.86 1.91 1.28 2.9 1.40
83.82 10.32 0.34 1.60 3.9 1.47
a Octylation reaction yield. b By difference. c With errors of
∼2.7% or less.
Table 3. 13C NMR Result and Calculated Number of
Octyl Groups n From Elemental Analysisb
Percentage of Aliphatic Carbonsa
(5)
Here a, b, and d are constants. Equation 5 could be rearranged
to eq 6:
∆V - d
) aC + b
C
Furrial
DM 153
DM 115
VLA 711
Hamaca
Cerro Negro
yielda mp range
(%)
(°C)
Furrial
DM 153
DM 115
VLA 711
Hamaca
Cerro Negro
octylated
asphaltenes
nb
62.4
75.0
76.5
71.8
75.4
73.4
50.0
61.8
71.0
59.8
65.9
63.5
3.3 ( 0.4
5.5 ( 0.7
3.1 ( 0.5
5.7 ( 0.6
5.3 ( 0.6
5.0 ( 0.6
a An error of 10% was assumed. b Calculated from elemental
analysis using eq 2.
Table 4. Measured (MALDI-TOF, 2100 LS) and
Calculated Molecular Weights and Other Results for
Asphaltenes and Octylated Asphaltene Samples
sample
Furrial
Furrial Oca
DM 153
DM 153 Oca
DM 115
DM 115 Oca
VLA 711
VLA 711 Oca
Hamaca
Hamaca Oca
Cerro Negro
Cerro Negro Oca
A1 Furrialb
Mn(ob)c Mw(ob)c Mn(calc)d
2040
2500
1720
2370
2020
2200
2240
2510
1840
2500
1700
2100
2000
3500
3770
2715
3620
3130
3640
3800
3690
2980
3810
2840
3264
3540
difference expected
(%)e
error (%)f
2600
-4
(22
2510
-5.6
(28
2540
-13.5
(20
3300
-24
(21
2655
-6
(28
2410
-12.87
(25
a Octylated asphaltene. b A1 fraction from Furrial asphaltene
isolated using the PNP method (see text). c Experimental (observed) values. d Calculated using eq 3. e Calculated using the
equation 100(Mn(ob) - Mn(calc))/Mn(calc). f Calculated using eq 4.
Contrary to asphaltenes, these octylated asphaltenes
dissolved easily in heptane at room temperature. Moreover, these derivatives have melting points of <105 °C,
and, in all cases, the octylation reaction yields were high
(see Table 2).
As shown in Table 3, the values of n range from 3 to
∼6. This means that at least three long octyl chains
1552
Energy & Fuels, Vol. 19, No. 4, 2005
Figure 1.
Acevedo et al.
13
C NMR spectra for DM 153 asphaltene samples: (top) asphaltene and (bottom) octylated asphaltene (OA).
were incorporated in the asphaltene molecule. The
presence of these chains leads to profound differences
in solubility and melting point, as shown in Table 2. As
is well-known, asphaltenes do not have melting points
and are insoluble in n-heptane and other paraffin
solvents, whereas the opposite in true for the octylated
derivatives. Because the presence of the octyl chains
must lead to a large reduction in the self-assembling
Molecular Weight of Petroleum Asphaltenes
Energy & Fuels, Vol. 19, No. 4, 2005 1553
Figure 2. Laser desorption-time-of-flight (LD-TOF) mass spectroscopy (MS) for (a) Furrial asphaltene and (b) octylated Furrial
asphaltene.
capacity of the sample, these changes are expected.
3.1. Mass Spectra Results. The LDI-TOF MS
results, measured at 2100 LS, are shown in Table 4. As
shown in this table, the Mn and Mw values of the asphaltenes leads to values of M
h n ) 1900 ( 200 and M
hw
) 3200 ( 400. In other words, these Mn and Mw values
are practically independent of sample type. Similar close
results were found for the OA derivatives (M
h n ) 2300
( 200 and M
h w ) 3600 ( 200). MW values for the OA
materials were higher than those for asphaltenes. For
the addition of n-octyl groups, this is the expected result.
Mn(OA), calculated using eq 3, were within 25% or
better of the observed results (see Table 4). Except in
the VLA case, these differences are within the expected
error for the calculation, which is an accumulation of
errors determining Mn(A) and n (see eq 4 and Table 4).
The Furrial A1 fraction was chosen because of its very
low solubility in toluene at room temperature (see
Experimental Section: Asphaltenes). However, no significant difference in the MW properties of the A1
fraction with the original sample (Furrial) was observed.
This interesting result will be discussed elsewhere.
The spectra corresponding to some asphaltene and OA
samples are shown in Figures 2-4. Some common
1554
Energy & Fuels, Vol. 19, No. 4, 2005
Acevedo et al.
Figure 3. LD-TOF MS for (a) DM 153 asphaltene and (b) octylated DM 153 asphaltene.
features could be observed in these spectra. Both
asphaltenes and OA samples span the same molecular
weight range (between ∼100 and 10 000). As described
previously, octylation leads to a relatively low-melting,
and more-soluble, sample than the asphaltenes. Thus,
it is expected that the volatilization of asphaltenes
would be enhanced by octylation. We found this to be
consistent with the MS measured at low power of 1900
LS. The case is illustrated in Figure 5 for the Furrial
samples, where the high-molecular-weight components,
which are clearly visible in the OA spectra (see Figure
5b), either are absent or have a lower intensity in the
asphaltene spectra (see Figure 5a).
The effect of laser power on the asphaltene and OA
spectra is shown in Figures 6 and 7, respectively. A
glance at these figures shows that a drastic change in
the MWD occurs between laser powers of 2100 and 2300
LS. At low power (1900 LS), a spectrum with poor
resolution and low intensities was obtained (see spectra
a in Figure 6; also see spectra a in Figure 5). At higher
power (2300 and 2500 LS, see spectra c and d in Figure
6; and 2300 LS, see spectra c in Figure 7), extensive
fragmentation is apparent from both spectra. The
change is very sharp for the OA Furrial case (compare
spectra b and c in Figure 7). As can be seen in these
spectra, as the power is increased, the fragmentation
shifts the MWD toward the low-molecular-weight values. From these findings, the laser power of 2100 LS
was selected as the best compromise and was used to
measure the values in Table 4.
Molecular Weight of Petroleum Asphaltenes
Energy & Fuels, Vol. 19, No. 4, 2005 1555
Figure 4. LD-TOF MS for (a) VLA 711 asphaltene and (b) octylated VLA 711 asphaltene.
As shown in Table 4, calculated Mn(OA) were reasonable similar to the experimental values. Percentage
differences were within the expected error in all cases.
This gives strong support to the laser power choice
previously mentioned. However, it should be mentioned
that differences between observed and calculated Mn
1556
Energy & Fuels, Vol. 19, No. 4, 2005
Acevedo et al.
Figure 5. Comparison of furrial asphaltenes spectra measured at a laser power of 1900 LS: (a) asphaltenes and (b) OA.
were always negative. This suggests that, even at this
relatively low power, some fragmentation is occurring.
As described previously, no matrix was required
because no significant differences were found between
Molecular Weight of Petroleum Asphaltenes
Energy & Fuels, Vol. 19, No. 4, 2005 1557
Figure 6. Power laser effect on the spectra of DM 153 asphaltenes. Laser power, from top to bottom, was as follows: 1900, 2100,
2300, and 2500 LS. Figure 2a is also included here (in Figure 6b), for comparison purposes.
the MALDI and the LDI (no matrix) spectra. A similar
finding has been recently reported.24
3.2.VPO Measurements. The VPO Mn values measured in this work are shown in Table 5. In all cases,
1558
Energy & Fuels, Vol. 19, No. 4, 2005
Acevedo et al.
Figure 7. Power laser effect on the spectra of furrial OA. Laser power, from top to bottom, was as follows: 2000, 2100, and 2300
LS.
Table 5. Comparison of Vapor Phase Osmometry (VPO)a
and Mass Spectroscopy (MS)b Number-Average
Molecular Weight (Mn) Values for Asphaltenes
Mn
values, suggesting that systematic errors were absent
(see the discussion section).
4. Discussion
sample
VPO
MS
difference (%)c
Furrial
DM 115
DM 153
Cerro Negro
Hamaca
VLA 711
2440
1400
1680
1760
2420
1720
2040
2020
1720
1700
1840
2240
16.4
-44.3
-2.4
3.4
24.0
-30.2
a In nitrobenzene at 100 °C. b Data taken from Table 4. c Calculated using the formula 100(Mn(VPO) - Mn(MS))/Mn(VPO). The
standard deviation of the differences was (27.0%.
fittings to eq 5 were excellent with r2 values equal or
very similar to 1. As shown in Table 5, comparison of
the two techniques afforded values which, on average,
are within (27.0% of each other. The Mn values in Table
5 are similar to other reported elsewhere using the same
technique and similar experimental conditions.1,2,14 The
differences in Table 5 gave both negative and positive
The good agreement between results from MS and
VPO and the comparison between observed and calculated Mn are good arguments to warrant the reliability
of the MS results summarized in Table 4. Thus, the
same molecular weight range found for both asphaltenes
and OA (100-10 000) and the comparison between
calculated and measured Mn(OA) values (see first and
third columns in Table 4) rules out any significant
impact of incomplete volatility or fragmentation on the
results measured at a laser power of 2100 LS.
The comparison in Figure 5 is consistent with the
higher volatility expected for OA materials. Figures 6
and 7 clearly indicate that fragmentation is an important issue when the laser power applied is too high.
According to these spectra, and the good comparison
between the observed and calculated Mn(OA) values in
Molecular Weight of Petroleum Asphaltenes
Energy & Fuels, Vol. 19, No. 4, 2005 1559
Figure 8. Comparison of (top) observed and (bottom) calculated MWD for the DM 153 asphaltene sample (see text).
Table 4, the power selected for these measurements
(2100) is a good compromise in the present case.
Similar molecular weight values and ranges were
reported in the previously cited paper by Tanaka et al.24
Thus, using a similar LDI MS technique, a molecular
weight range of 200-8000 amu and molecular weight
averages of ∼1650 were reported by these authors for
Maya asphaltenes. This value is close to those given in
Table 4.
Of course, a significant increase in molecular weight
averages should be expected if polymerization has any
importance in the present measurements. As found by
Tanaka et al., using low-molecular-weight (∼300) model
compounds, polymerization was promoted in this case
by increasing the laser power.24 However, no such
polymerization was detected when either a high-molecular-weight (∼700) aromatic model or asphaltenes were
used.24
In agreement with these findings, increasing the laser
power gave no evidence of polymerization in the present
case. Instead, when power was increased, the MWD for
both asphaltenes and OA shifted to lower-molecularweight values (see Figures 6 and 7). Thus, based on
these results, polymerization could be ruled out.
The previously mentioned conclusions are reinforced
by the comparison with the VPO results (see Table 5).
The differences between the Mn values measured using
the two techniques were determined to be within an
average standard deviation of 27%. In view of the
expected errors for these measurements (between ∼10%
and 20%), the agreement is very good. Differences
between the two techniques are either negative or
positive, strongly suggesting that these are not systematic. Thus, systematic asphaltene aggregation in VPO
or extensive fragmentation (at a laser power of 2100 LS)
and polymerization in MS could be discarded.
Similar VPO Mn values for 14 asphaltene samples
were reported some time ago for both nitrobenzene (Mn
) 1900 ( 300, T ) 100 °C) and pyridine (Mn ) 2770 (
490, T ) 37 °C).1 Values of ∼2000 (pyridine, T ) 50 °C)
were reported by other researchers.19 When measured
in other polar solvents, such as THF16 or dibromomethane,1 higher values are usually obtained, presumably because of aggregation promoted by lower
solvent polarity and lower temperature.
The previously mentioned small standard deviations
found for the reported VPO values in pyridine,1 for the
VPO values in Table 5, and for the average values in
Table 4 are worthy of comment at this point. It suggests
that, when molecular weight values are determined
using adequate conditions, relatively small differences
in the average molecular weight values will be observed.
Figure 8 is an example of how well the experimental
MWD could be reproduced by a LND. The comparison
is for the DM 153 sample. The only imputed data for
the calculation were the experimental Mn and Mw
values. The details of the calculated curve are given in
the Appendix. The resemblance is very good. Using SEC,
similar good fitting to LND has been reported for
asphaltenes3 and OA materials.15 As mentioned in the
1560
Energy & Fuels, Vol. 19, No. 4, 2005
Introduction section, these fittings could be expected for
a natural distribution of compounds. The good comparison shown in Figure 8 would not be expected in the case
of extensive fragmentation or polymerization during the
MS experiment.
5. Conclusions
As shown previously, the selection of the proper
conditions for the laser desorption ionization (LDI) mass
spectroscopy (MS) measurements is very important to
obtain meaningful results. In this sense, a compromise
between volatility and fragmentation must be reached.
Using a power that is too low to avoid fragmentation
leads to poor spectra, lacking the less-volatile, highmolecular-weight components. On the other hand, the
use of a power that is too strong produces extensive
fragmentation, leading to drastic changes in the molecular weight distribution (MWD). The molecular weight
averages and ranges observed for asphaltenes in this
work, using an intermediate laser power, were validated
(i) by comparison with the observed and calculated
molecular weight properties of the octylated asphaltene
derivative (OA) materials, (ii) by comparison with VPO
values, both from this work and with literature values,
and (iii) by comparison with reported LDI MS performed
under similar conditions. According to these comparisons and the previous discussion, these values are not
significantly affected by artifacts such as polymerization, fragmentation, or incomplete volatilization.
Acknowledgment. We would like give our deep
appreciation to the following persons and institutions:
Drs. Daniel Jones and Jonathan Mathews (Pennsylvania State University), for preliminary LD MS measurements; Drs. Marı́a Antonieta Lorente and Olga Leon,
for the oil samples; Lic. Javier Sayago, for collecting
some of the oil samples from the field; and Dr. Jacques
Jose (University Claude Bernard-Lyon). One of the
authors (S.A.) is gratefully to the University of Pau for
Acevedo et al.
a visiting professorship, when part of this study was
conducted, and to the Post-graduated Cooperation Program between France and Venezuela. Also our thanks
to Lic. Betilde Segovia and to FONACIT (under Grant
Nos. G97000722 and CONICIT-CONIPET 97004022
and CDCH Grant No. 03.12.4338/99) for financial support.
Appendix
The normalized frequency f in a log-normal distribution (LND) is given by eq A-1:3,21,26
f)
[ (
)]
ln M - ln M0
1
exp -0.5
σ
σx2π
2
(A-1)
Equation A-1 gives the distribution of mass in terms of
ln M.
To apply this equation, we must know the values of
M0 and σ. These are given by eqs A-2 and A-3:3
M0 ) xMnMw
(A-2)
x( )
(A-3)
σ)
ln
Mw
Mn
Using the DM 153 results (see Table 4), values of M0 )
2161 and σ ) 0.675 were obtained. For comparison with
the MS distribution, eq A-1 must be divided by M, to
covert the mass distribution to a molar distribution.
Thus, we must consider the function f/M, rather than f.
Moreover, for the same purposes of comparison, we
plotted the function 105(f/M) + 10. In this way, both the
experimental and calculated distribution were placed
in the same scale. The comparison is shown in Figure
8.
EF040071+
(26) Hiemenz, P.; Rajagolapan, R. Principles of Colloid and Surface
Chemistry, Third Edition, Revised and Expanded; Marcel Dekker: New
York, 1997; Appendix C, p 636.